Local anaesthetic toxicity

This chapter tries to address Section K2(ii) of the 2017 CICM Primary Syllabus, which expects the exam candidate to demonstrate an "understanding of the pharmacology of local anaesthetic drugs, including their toxicity". Because the majority of past paper questions had specifically wanted to discuss local anaesthetic toxicity, it clearly deserved its own page. 

• Question 19 from the second paper of 2018
• Question 22 from the second paper of 2012
• Question 3 from the first paper of 2009

Additionally, this has recently crossed over into Second Part Exam territory, as a 25% fraction of Question 3 from the first paper of 2022.

In summary,

At the time of writing, and hopefully at the time of reading,  Finucane's  Complications of regional anesthesia (2006) is available for free somehow, and the local anaesthetic toxicity chapter by Brown from this textbook was probably the best-referenced and most accessible. Naguib et al (1998) also appears to be free, and is remarkably comprehensive. Together, these resources could facilitate a true deep dive into local anaesthetic toxicity.

Central nervous system toxicity from local anaesthetic agents

At this stage, the reader may naturally be wondering: if this elegant mechanism of action numbs the nerves and silences neurotransmission in the peripheral nervous system, then surely it would also work its magic if one were to inject it directly into the brain? In short, what CNS effect can we expect with big local anaesthetics given systemically? 

Well, seizures. Seizures is the answer. There is a dose-dependent and well-recognised syndrome of CNS toxicity from local anaesthetics which is excitatory rather than depressant in its net effect. Sure, sedation would probably be seen, just as it is with general anaesthetics, except that inhibitory neuronal activity is preferentially suppressed first by systemic local anaesthetics.

We know a lot of this because of cruel human and animal experiments. Scott(1986) recalls some of their own work from the seventies, when human volunteers were administered lignocaine until over toxicity was achieved. The author was surprised by the unpredictable and erratic phenomenology of local anaesthetic overdose. Each patient seems to have experienced a different set of symptoms:

"For example, some would report tinnitus, others not; some would talk irrationally and even become unconscious for a few seconds, while others remained lucid even in the presence of widespread muscular twitching."

The clinical effects of CNS toxicity generally begin as perioral paresthesia and weird visual disturbances. "Objects in the visual fields appear to oscillate either from side to side or up and down, or both. The subject may try to follow these movements with the eyes and may be diagnosed as having nystagmus, although this is incorrect as the eyes can be held steady if the subject concentrates", i.e. this oscillating weirdness was entirely in their mind. Later, lightheadedness and tinnitus would come. The subjects would ask to lay down, but they just could not keep still. Motor overactivity like shivering, muscle twitching and tremors was accompanied by more disturbing features like slurred speech and incoherent rambling. The next step was generalised tonic-clonic seizures, though Scott hastened to add that none of the volunteers were taken to this point. Shibata explored this dose range in EEG-monitored cats, getting infusions of lignocaine at a dose rate of up to 15mg/kg, and were clearly able to demonstrate a dose-dependent epileptiform effect.

With increasing doses, excitatory neurotransmission is also affected, and a profound CNS depression occurs. This is not the sort of toxicity one might see with accidental administration of peripheral venous lignocaine, as the CNS concentration required would be truly preposterous. One may theoretically see it when a hilarious factor-of-ten dilution error is perpetrated, or where somebody accidentally injects concentrated local directly into the arterial circulation of the brain. That is the setting of some sheep experiments by Ladd et al (2002), who were able to produce something resembling burst suppression with heroic doses of intracarotid bupivacaine. 

Cardiovascular toxicity from local anaesthetic agents

The central nervous system is much more sensitive to local anaesthetic toxicity than the cardiovascular system presumably because of different affinity of the agents for the cardiac version of voltage-gated sodium channels. According to D.L. Brown (2006, p.65), three times as much lignocaine is required. However, often the patient may already be anaesthetised and therefore less inclined to report perioral numbness. If one has accidentally used an unreasonable amount of local anaesthetic, the following haemodynamic features would be observed as the toxic doses increases:

• Sympathetic activation at first, with tachycardia and hypertension
• Then, myocardial depression, moderate hypertension, and decreased cardiac output.
• Finally, "peripheral vasodilatation, profound hypotension, myocardial conduction abnormalities, sinus bradycardia, ventricular arrhythmias, and ultimately cardiovascular collapse". 

The mechanisms of these changes are thought to be due to the inhibition of sodium channels in the cardiac conduction system, though as we have already discussed there's a lot of other channels being blocked by local anaesthetics and there is really no published material out there exploring the cardiovascular implications of this. Observed effects include:

• Bradycardia is observed, as conduction time is prolonged, and spontaneous pacemaker activity is prolonged.
• The QT interval should be shortened
• Contractility will be depressed with higher doses
• Cardiac arrest may occur with large bolus doses.
• A decreased maximal rate of depolarization (Vmax) of Purkinje fibres and cardiac myocytes is seen, which results in the prolongation of Phase 0 and therefore an increase in the width of the QRS complex (analogous to what you see in tricyclic overdose). Yes, I hear you say, lignocaine is an Class Ib antiarrhythmic and Class Ib drugs are not supposed to do anything to the width of the QRS. But that is when they are used responsibly, and in conservative doses, and that's lignocaine, the most cardiogentle of them all. When Mauch et al (2010) intentionally overdosed some neonatal pigs with nasty cardiotoxic bupivacaine, QRS complexes definitely widened, and impressive T wave changes were observed. Finally, with truly absurd doses, the rhythm degenerates into this:
  
  That is the ECG of a 9-month-old girl from a case report by Froehle et al (2012), who had accidentally received a  20mg/kg bolus of mepivacaine, and was successfully rescued by ECMO.

Vasodilation and?...or vasoconstriction may occur. At MAC95.com, frustrated authors point out that two of the most influential textbooks for the ANZCA primary offer contradictory statements on this topic. According to Hemmings (2013), they vasoconstrict at low doses and vasodilate at high doses, but according to Peck & Hill they vasodilate at low doses and vasoconstrict at high doses. To act as tiebreakers in this battle of titans, one can offer Blair (1975) and Brown (2006), who both back Hemmings:

"Typically, at low concentrations local anesthetics may cause increased tone in vascular beds, whereas at higher concentrations they produce a decrease in vascular tone. At extremely high blood levels, there is profound peripheral dilatation because of a direct relaxing effect on vascular smooth muscle in almost all beds."

That vasodilator response appears to be in part due to the sodium channel effect of local anaesthetics acting directly on the sympathetic innervation of vascular smooth muscle, and in part due to some completely unrelated interaction between the drug and the nitric oxide vasodilator system. Newton et al (2007) were able to reverse 60% of this vasodilator effect with a nitric oxide synthase inhibitor.

The influence of drug factors on local anaesthetic toxicity

• Choice of agent matters; bupivacaine is thought to have greater potential for cardiotoxicity, whereas all the other agents are apparently less cardiotoxic. This is usually expressed in terms of the "CC/CNS ratio", i.e. the difference in dose required to cause cardiac complications vs. central nervous system toxicity. As discussed, CNS toxicity usually requires about three times less local anaesthetic than cardiovascular toxicity. This empirically derived ratio is as high as 7.1 for lignocaine, and as low as 2.4 for bupivacaine (Garg et al, 2020).
• Dose of the drug:  it goes without saying that an unreasonably large dose will give rise to toxicity. One needs to remember that this could happen accidentally if dose calculations to an actual body weight are performed in obese or pregnant patients
• Site of administration: Large volume blocks anywhere, and small volume blocks in the head and neck, are most susceptible because of the risk of accidental large-volume intravenous injection (eg. into the external jugular), or because systemic absorption is particularly good if a very large pool of drug has collected. Moreover, some sites are more associated with systemic absorption than others (which has to do with their vascularity as well as the proximity of large vascular structures).  These sites are:
  • Increased risk of direct intravascular injection:
    • Interscalene block
    • Brachial plexus block
    • Stellate ganglion block
    • Intercostal nerve block
  • Increased risk of rapid absorption:
    • Scalp
    • Bronchial mucosa
    • Interpleural cavity
    • Epidural
• Drug interactions can increase or decrease the risk of local anaesthetic toxicity. Classically, the concomitant use of adrenaline as a vasoconstrictor in subcutaneous infiltration results in decreased systemic absorption and therefore the possibility of using a higher maximum dose. Logically, that means the concurrent use of a vasodilator (eg. GTN infusion) can produce an increase in risk. Similarly, ephedrine or other inotropes could produce an increase in tissue perfusion and speed up the washout of the drug into the systemic circulation. Some drugs can also compete for protein binding with local anaesthetics: for example, phenytoin, quinidine and desipramine can displace bupivacaine from plasma proteins. Lasly, interference with hepatic metabolism of amide anaesthetics (eg. by cimetidine) or with hydrolysis of esters (eg. ecothiopate) could result in delayed clearance. There's actually a massive amount of even rarer drug interactions that could be mentioned (eg. where flumazenil lowers the seizure threshold and potentiates the CNS toxicity), but rather than listing them all here, one could probably just offer this excellent article by Naguib et al (1998) as a reference.

The influence of patient factors on local anaesthetic toxicity

Christie et al (2015) lists some of the patient factors which influence local anaesthetic toxicity, which the examiners complained were "often omitted" from Question 19 from the second paper of 2018. These are:

• Acidosis: Only the charged form of a local anaesthetic molecule can bind to a voltage-gated sodium channel, and in the presence of intracellular acidosis, more of them will be present in this active ionised state. Or, acidosis changes the protein binding and makes more of the agent available to be free and do crimes. Or, acidosis increases the partition coefficient of local anaesthetic into the myocardium. All of these possible theories are quoted as explanations of the observed phenomenon, that acidosis makes the haemodynamic consequences of local anaesthtic toxicity much worse. Nancarrow and colleagues destroyed a large number of sheep to debunk this principle in 1987, but it still persists in textbooks. It may well be that coexisting acidosis on its own has enough of a cardiodepressant effect to give people this impression.
• Old age: the elderly have less hepatic blood flow and a thousand other reasons to clear all drugs much more slowly, therefore being at risk of accumulated doses.
• Young age: Children and neonates have a lower α₁-acid glycoprotein level, which means for any given total blood concentration of local anaesthetic, the free fraction will be greater.
• Pregnant patients also have a lower α₁-acid glycoprotein level, and beyond that their pregnancy-induced hyperaemia and increased cardiac output increases the washout of local agent from its nerve-block depot, and into the systemic circulation.  
• Needless to say, the concurrent use of another potent antiarrhythmic would cause cardiovascular trouble earlier and with less warning.
• Hyperkalemia potentiates the cardiotoxic effects of local anaesthetics. Avery et al (1984) determined that, at a potassium of only 5.4 mmol/L, the dose of local anaesthetic required to cause severe cardiovascular effects was halved, as compared to a potassium of 2.7 mmol/L. The same has been observed in various case reports, where a hyperkalemic and acidotic patient surprised the anaesthetist by developing local anaesthetic toxicity at an unexpectedly low dose of the agent. The authors did not speculate on what the mechanism of this might be. 

Other toxic effects of local anaesthetics

Though CNS and cardiovascular effects are the most important, one may be able to score a couple of extra marks by mentioning some of the more outré side effects, which one does not really see in clinical practice. These include:

• Methaemoglobinaemia with prilocaine, because prilocaine is metabolised into aminophenol metabolites ortho-toluidine and N-propylalanine, both of which can oxidize haemoglobin into methaemoglobin,
• Myonecrosis (with direct intramuscular injection)
• Allergic reaction (with ester agents, as they are metabolised to produce para-aminobenzoic acid, and this stuff can be immunogenic, promoting anything from a rash to anaphylaxis). There's also a very real non-theoretical allegenic potential of methylparaben used as preservative excipient, but as Brown (2006) points out, this is an incredibly common food additive and it would be extremely unlikely for any individual in the developed world to go through life without developing immune tolerance of it (which is why it still ends up being added to these ampoules). 

Management of local anaesthetic toxicity

Supportive management is fairly basic, and consists of doing empirical things which counteract the toxicities. Patient having seizures? Give a benzodiazepine to raise the seizure threshold. Patient lost their airway? intubate them. Patient suffered total cardiovascular collapse? Support them with ECMO. It would be unexpected for these predictable answers to be rewarded with marks in the CICM exams. More likely, the examiners would be more interested in the specific management of local anaesthetic toxicity. It is interesting because its goal is to change the availability of the free fraction of local anaesthetic, thus making it less available.

• Decrease the availability of circulating local anaesthetic agent:
  • Alkalinise or hyperventilate: the agent's binding to plasma proteins is pH-dependent, and more of the agent will be protein-bound under alkaline conditions. This also decreases the concentration of intracellular ionised agent, which is the active form capable of binding to the 
  • Lipid infusion to alter the distribution: This is usually intralipid, a bolus of 1.5 mL/kg IV over 1 minute, followed by a continuous infusion 0.25 mL/kg/minute.

Rationale for lipid infusion in local anaesthetic toxicity is to remove the highly lipophilic local anaesthetic from the circulation, and to confine it to the lipid emulsion. This allows the agent to be cleared more slowly, and decreases the availability of the free fraction. Ok et al (2018) is an excellent summary of these mechanisms, which are presented below in expanded point form:

• Lipid sink: the highly lipid-soluble local anaesthetic molecules are absorbed into the lipid emulsion droplets, which decreases the free fraction of the drug in the circulation
• Tissue extraction: because the free fraction in the circulation drops, redistribution from target tissues (CNS, myocardium) will occur, reducing toxicity in those organs
• Lipid shuttle: the fatty droplets of lipid emulsion act as a carrier which delivers the local anaesthetic to the liver, enhancing the rate of elimination (apparently this is also referred to as a "lipid subway")
• Metabolic changes in the myocardium:  the increased fatty acid supply reverses local-anaesthetic-induced reduction in fatty acid metabolism in the cardiac mitochondria
• Inoconstrictor effects though the inhibition of nitric oxide release and some positive inotropic effects, which appears to be an intrinsic property of the lipid emulsion
• Reversal of cardiac sodium channel blockade by a mechanism apparently related to fatty acid-mediated modulation of cardiac sodium channels